| 201 | 0 | 330 |
| 下载次数 | 被引频次 | 阅读次数 |
为提高抄表效率和读数精度，针对当前字轮式水表读数精度低、参数量大等问题，提出一种基于改进YOLOv8n的字轮式水表数字识别算法。通过引入GhostConv和注意力机制HDCA(High-Resolution Dual-Channel Attention)设计新的GDC2f模块，简化特征提取网络，并增强模型对水表字符的提取能力；采用GSConv代替原有的卷积核，同时引入Slim-Neck特征融合网络，增强模型对小目标特征的表达能力，降低模型的参数量；采用MPDIoU损失函数，提高边界框的定位能力和模型的收敛速度。实验结果表明，改进模型的精确度、召回率和平均精度分别提高了1.3%、2.4%和3.3%,计算量、参数量和模型大小分别减小了2.9 GB、0.79×106和0.5 MB。
Abstract：In order to improve the efficiency and accuracy of meter readings for word-wheel water meters, a digital recognition algorithm for word-wheel water meters based on improved YOLOv8n is proposed to address the problems of low reading accuracy and excessive parameters. Firstly, a novel GDC2f module is designed by introducing GhostConv and the attention mechanism HDCA(High-Resolution Dual-Channel Attention). This design simplifies the feature extraction network while enhancing the model's ability to extract water meter characters. Secondly, GSConv is employed in place of the original Conv, and a Slim-Neck feature fusion network is introduced, which enhance the feature expression ability for small targets and reduce the number of parameters. Finally, the MPDIoU is adopted to optimize the model, improving the ability of the bounding box localization and the convergence speed of the model. Experimental results show that the improved model increases the precision, recall and average accuracy by 1.3%, 2.4% and 3.3%, respectively. It also decreases the computation, parameters and model size by 2.9 GB, 0.79×106 and 0.5 MB, respectively.
[1]王一民，钱浩涵，王佩君，等.智能水表的应用及产业发展现状分析[J].计量与测试技术，2022,49(3):102-105,108.
[2]XIE W,LIU W,DAI Y,et al.Multiscale attention and feature decomposition network for surveillance vehicle detection[J].IEEE Sensors Journal,2024,24(19):31573-31588.
[3]GIRSHICK R,DONAHUE J,DARRELL T,et al.Rich feature hierarchies for accurate object detection and semantic segmentation [C]//2014 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Columbus:IEEE,2014:580-587.
[4]GIRSHICK R.Fast R-CNN [C]//2015 IEEE International Conference on Computer Vision(ICCV).Santiago:IEEE,2016:1440-1448.
[5]REN S Q,HE K M,GIRSHICK R,et al.Faster R-CNN:Towards real-time object detection with region proposal networks[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Boston:IEEE,2015:91-99.
[6]REDOMN J,FARHADI A .YOLOv3:An incremental improvement [EB/OL].(2018-04-08) [2024-11-05].https://arxiv.org/abs/1804.02767.
[7]BOCHKOVSKIY A,WANG C,LIAO H.YOLOv4:Optimal speed and accuracy of object detection [EB/OL].(2020-04-23) [2024-11-05].https://arxiv.org/abs/2004.10934.
[8] LIU W,ANGUELOV D,ERHAN D,et al.SSD:Single shot multibox detector[C]// European Conference on Computer Vision.Cham:Springer,2016:21-37.
[9]孙顺远，杨挺.基于深度学习的仪表目标检测算法[J].仪表技术与传感器，2021(6):104-108.
[10]李少波.基于深度学习的水表读数识别应用研究[D].南昌：南昌大学，2021.
[11]翟娅娅，朱磊，张博.基于改进YOLOv4网络的水表读数识别方法[J].科学技术与工程，2022,22(21):9207-9214.
[12]汤鹏，刘毅，魏宏光，等.基于Mask-RCNN海上升压站数字式仪表读数的自动识别算法[J].红外与激光工程，2021,50(S2):163-170.
[13]陈文萍，娄嘉骏，江少锋，等.基于YOLOv5的水表字轮读数自动识别方法[J].仪表技术，2023(3):43-46,74.
[14]高嘉，刘涛，王显峰，等.TWRD-Net:一种用于曳引钢丝绳缺陷的实时检测网络算法[J].仪器仪表学报，2023,44(6):223-235.
[15]PANIGRAHY S,KARMAKAR S.Real-time condition monitoring of transmission line insulators using the YOLO object detection model with a UAV[J].IEEE Transactions on Instrumentation and Measurement,2024,73:1-9.
[16]LIU C,WEI S,ZHONG S,et al.YOLO-powerLite:A lightweight YOLO model for transmission line abnormal target detection[J].IEEE Access,2024,12:105004-105015.
[17]BODLA N,SINGH B,CHELLAPPA R,et al.Soft-NMS:Improving object detection with one line of code[C]// 2017 IEEE International Conference on Computer Vision(ICCV).Venice:IEEE,2017:5561-5569.
[18]LIN T Y,DOLLAR P,GIRSHICK R,et al.Feature pyramid networks for object detection[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Honolulu:IEEE,2017:936-944.
[19]LIU S,QI L,QIN H,et al.Path aggregation network for instance segmentation[C]// 2018 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Salt Lake City:IEEE,2018:8759-8768.
[20]HAN K,WANG Y,TIAN Q,et al.Ghostnet:More features from cheap operations[C]// 2020 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Seattle:IEEE,2020:1577-1586.
[21] LI H,LI J,WEI H,et al.Slim-neck by GSConv:A better design paradigm of detector architectures for autonomous vehicles [EB/OL].(2022-06-06) [2024-11-05].https://arxiv.org/abs/2206.02424.
[22] MA S,XU Y.Mpdiou:A loss for efficient and accurate bounding box regression [EB/OL].(2022-07-14) [2024-11-05].https://arxiv.org/abs/2307.07662.
基本信息:
DOI：10.13367/j.cnki.sdgc.2026.01.012
中图分类号:TP183;TP391.41;TU821.2;TU991.63
引用信息:
[1]乔世超,袁玉英,齐瑞洁.基于改进YOLOv8n的字轮式水表数字识别算法[J].山东理工大学学报(自然科学版),2026,40(01):1-8.DOI:10.13367/j.cnki.sdgc.2026.01.012.
基金信息:
淄博市科技型中小企业创新能力提升工程项目(2023tsgc0043); 张店区校城融合发展计划项目(2021JSCG0018)